Automated MySQL failover with MHA: getting started & moving past its quirks

In the third edition (2012) of High Performance MySQL, MHA usage is described as: "anyone other than Yoshinori who is using it in production, and we haven't used it ourselves." Little known to the authors, we've spent quite a lot of time working on MHA and figuring out its kinks. It has also been deployed in many a production environment over the past year. Today MHA is in use in places like DeNA (gaming), Facebook (social networking), SK Telecom (telecoms), and DAPA (defense).

Learn:

* the potential pitfalls of automated failover and their workarounds

* how regular MySQL replication (asynchronous or semi-synchronous) can give you high availability with automated failover capabilities

* dealing with the semi-sync plugin, and understanding what you can get with enhanced semi-sync

* about virtual IP failover & integration with Pacemaker+Corosync

* if using the mysql-utilities tool, mysqlfailover, makes any sense

* knowing when to choose between PRM and MHA

* using MHA with GTID implementations in MySQL 5.6 & MariaDB 10

Learn from the experience in deploying and doing further engineering glue around the solution that is MHA. At the end of the talk, the attendee will be able to download MHA, and deploy it in an environment with a minimum of 3-nodes with ease and think about further integration to build their own cloud architectures.